Is there an analytic theory of automorphic functions for complex algebraic curves? (Q2185859)

Langlands has been asserting that there should exist an analytic theory of automorphic functions for geometric Langlands theory of complex algebraic curves. Motivated by his insistence, the paper under review outlines an approach, which is detailed in the work of the author et al. [``An analytic version of the Langlands correspondence for complex curves'' Preprint, \url{arXiv:1908.09677}]. The Langlands program was initiated by Langlands in the late 1960s. Let \(F\) be a number field, meaning a finite extension of \(\mathbb{Q}\), or a function field, meaning the field of rational functions on a smooth proper curve \(C\) over a field field. Let \(G\) be a reductive group over \(F\). A protagonist of the program is automorphic forms, certain functions on \(G(F)\backslash G(\mathbb{A}_F)/K\) where \(\mathbb{A}_F\) is the ring of adeles of \(F\) and \(K\) is a compact subgroup of \(G(\mathbb{A}_F)\). One can find a commutative algebra of Hecke operators acting on automorphic forms. Then a crucial insight of Langlands was to study their spectral decomposition and describe it in terms of, roughly speaking, homomorphisms from the Galois group \(\text{Gal}(\overline{F}/F)\) to the Langlands dual group \(\check{G}\). Then from the early 80s, the geometric Langlands program was developed by Drinfeld and Laumon where one shifts interest from functions on the double coset to sheaves on a certain algebraic stack (for example, the moduli stack \(\text{Bun}_G\) of principal \(G\)-bundles on the curve \(C\) in the unramified case): in the case of function fields, one can recover the original formulation of Langlands if Grothendieck's function-sheaf correspondence is applied. An important point of departure from the initial formulation is that the sheaf-theoretic formulation makes sense even for a complex algebraic curve where there is no function-sheaf correspondence. In the complex setting, Beilinson and Drinfeld reformulated the crux of the program as a spectral decomposition of Hecke functors on the (derived) category of sheaves on \(\text{Bun}_G\). In particular, the entire framework was completely algebraic and there didn't seem to be a room to find an analytic theory as for the classical setting. Recently, Langlands made a proposal for an analytic theory of spectral decomposition of Hecke operators for complex curves in his paper ``On the analytical form of the geometric theory of automorphic forms'' (written in Russian). In the paper under review, Frenkel identifies some issues in Langlands' proposal with providing explicit and concrete examples and suggests a different analytic theory of automorphic forms based on the joint work with Etingof and Kazhdan. The main suggestion is to study spectral decomposition properties of a commutative algebra different from the one of Hecke operators. To identify the commutative algebra, let \(K\) be the canonical line bundle on \(\text{Bun}_G\). Consider the algebra of holomorphic differential operators on \(K^{1/2}\), its complex conjugate acting on \(\overline{K}^{1/2}\), and their tensor product. This is the algebra of interest which will play the role analogous to the one of Hecke operators. Namely, the author notes that the algebra naturally acts on a certain Hilbert space of sections of the line bundle \(K^{1/2}\otimes \overline{K}^{1/2}\) over \(\text{Bun}_G\) and studies their spectral decomposition property. It is worth mentioning that this spectral problem was considered earlier by Teschner in ``Quantisation conditions of the quantum Hitchin system and the real geometric Langlands correspondence''. One may ask why it is reasonable to think of the given problem as the spectral problem relevant for the geometric Langlands program. The author provides an answer in Remark 4.1; he recalls what happens to automorphic forms of \(G\) over a number field \(F\) where one has to study representations of \(G(\mathbb{Q}_p)\) and \(G(\mathbb{C})\). That is, while the spherical Hecke algebra comes in for representation theory of \(G(\mathbb{Q}_p)\), one uses the center \(Z(\mathfrak{g})\) of the universal enveloping algebra \(U(\mathfrak{g})\) for representation theory of \(G(\mathbb{C})\). In the geometric setting where we would discuss representation theory of \(G(\mathbb{C}(\!(t)\!))\), this leads us to the study of the center of \(U(\mathfrak{g}(\!(t)\!))\) or its variant. In fact, the earlier work of the author with Feigin ``Affine Kac--Moody algebras at the critical level and Gelfand--Dikii algebras'' identifies such a center at the critical level, which is very closely related to the global differential operators on \(K^{1/2}\). The paper ends with formulating a concrete conjecture for the spectral problem. To identify it, recall that Beilinson and Drinfeld (in the unpublished book ``Quantization of Hitchin's integrable system and Hecke eigensheaves'') identified the algebra of differential operators on \(K^{1/2}\) in terms of the algebra of functions on the space of \(\check{G}\)-opers, based on the work of Feigin and the author mentioned above. With this identification, the spectral problem conjecturally leads to the study of opers whose monodromy takes values in the split real form of \(\check{G}\). The paper ends with the example of \(G=\text{GL}_1\) and \(C=E_i=\mathbb{C}/(\mathbb{Z} + \mathbb{Z}i)\) to provide a simple illustration of the conjecture.